Upgrade failed due to unclean database shutdown II


(Andreas Motl) #1

Hi there,

first things first: Thanks for your excellent work on Discourse. We are running an instance on http://community.hiveeyes.org/ and choosing Discourse was one of the best things which could happen to us. Keep up the spirit!

While running this forum, we had one minor hiccup the other day with Discourse stuck on "Currently Upgrading" we have been able to resolve quickly.

A few minutes ago, we had another one when attempting to upgrade Discourse using launcher rebuild app on a slightly loaded machine: We experienced the very same behavior as @ljpp had with Upgrade failed due to unclean database shutdown.

Cheers,
Andreas.


(Andreas Motl) #2

While the rebuild was stuck with ** FAILED TO BOOTSTRAP ** following the failed Pups::ExecError due to the unclean postgres shutdown, we have been able to start the instance again (after that, all the chrome re. customized header etc. was missing), followed by a clean shutdown and a successful rebuild:

launcher start app
launcher stop app
launcher rebuild app

Everything now works as intended. Thanks for listening.


(Andreas Motl) #3

Now, we are wondering whether it would be possible that the pgctl -w thing mentioned at

currently would not apply in rebuild situations without a preliminary stop. At least, Discourse seems to have recovered its Postgres database after a regular start, right?